Methods for Brain Connectivity Analysis with Applications to Rat Local Field Potential Recordings

Area of Science:

• Neuroscience
• Computational Neuroscience
• Data Science

Background:

• Understanding brain dependence networks is crucial for deciphering neural mechanisms of perception, action, and memory.
• Hippocampal local field potential (LFP) data provides insights into neural dynamics during information processing.

Purpose of the Study:

• To present and evaluate a range of statistical methods for analyzing brain network dependence.
• To investigate the encoding of nonspatial olfactory information in rat hippocampus using multivariate LFP time series.
• To compare the strengths and limitations of classical and advanced techniques in capturing neural dynamics and connectivity.

Main Methods:

• Exploratory techniques: correlation, partial correlation, spectral matrices, coherence.
• Advanced methods: Granger causality (GC), robust canonical coherence analysis, spectral transfer entropy (STE), wavelet coherence.
• Emerging approaches: Topological Data Analysis (TDA), deep learning-based canonical correlation frameworks.

Main Results:

• Foundational connectivity insights were established using exploratory techniques.
• Dynamic and nonlinear interactions were captured by advanced methods like GC and STE.
• TDA and deep learning frameworks showed potential for multi-scale feature extraction and connectivity modeling.

Conclusions:

• A comprehensive comparison of state-of-the-art techniques for brain dependence network analysis was provided.
• The study highlights the unique strengths of various methodologies and addresses computational challenges.
• This work paves the way for future research in neural connectivity and information processing.